The Agentic Age: Intelligence, Causality, & Risk

Agentic AI promises to bring the inferential power of neural networks into the space of autonomous problem solving systems. Agentic AIs are a feral technology, having started as a domesticated tool, they are projected to become a wild, self-sustaining, population. A population that will grow into markets, institutions, and homes.

The current interest in Agentic AI, namely the abstract problem that agents, markets, scientists, and investors are all trying to solve is ultimately solving policy search under uncertainty in structured environments. Whether one is searching for new strategies, products, or markets, all require a means of exploring an open-ended space in conditions of bounded rationality. The interesting question for AI agents is whether they can generate genuine novelty and explore regions of policy space that were not implicit in prior training data or merely sample from a pre-given space with high coverage.

In this meeting we shall explore the wilds of agentic AI exploring both costs and benefits of a feral digital intelligence.